内容分发策略是影响一个内容分发网络(CDN)性能的主要原因之一。然而最早使用的各种启发式算法已经不能满足用户对时延的要求,因此许多研究人员开始利用数学规划的手段解决这一难题。但是,目前所提出的数学模型都存在其局限性和不足之处。本文从各个实际角度出发,考虑了流媒体速率、文件维持消耗、网络拓扑等各种情况,并对各种情况建立了完善的数学模型。同时,本文还利用节点来表示一个区域内的多个服务器,从而降低变量空间。
A major factor that heavily affects the performance of a Content Delivery Network (CDN) is content placement scheduling. The heuristic algorithms used in the past few years are unable to meet the delay requirements of users, many researchers have begun to use mathematical programming means to solve this dilemma. But the recent mathematical models have their limitations and inadequateness. From the practical point of view, we consider the rate of streaming media, documents maintenance consumption, network topology and other conditions, and establish a perfect model for each circumstance. Meanwhile,we use a node to represent multiple servers, which are geographically close to each other, thus reducing the variable space.